Wired recently published an interview with a library science researcher in which they asked about the current state of scientific literature (hard and social sciences), how we got here, and why that is a problem for researchers, students, librarians, and everyone else.
It's very readable and worth a couple of minutes.
People will often ask us why we don't have a subscription to X journal. Most libraries would love to have a subscription to everything, but for the vast majority of us, it's just not feasible. As the cost of the "must-have" titles ratchets ever skyward, libraries around the world are forced to either expand their budgets drastically or to make difficult decisions regarding their "nice-to-have" subscriptions.
The balancing act for scholarly publishing today is in finding a model that rewards good researchers with the credit they deserve while weeding out the poorly-done papers. Then, access costs need to be affordable enough for institutions and individuals to actually see that good research and give it the wider audience it deserves.
Hopefully open-access publishing will continue to grow, improve, and gain prestige. Trailblazing journals like PLOS ONE are demonstrating that "free to read" can also be "high-quality" and "impressive on a CV."